Seven agencies across four countries tied fake recruiters and take-home coding tasks to eight months of infections and drained cryptocurrency wallets.
Reality
- Evidence62
- Adoption70
- Hype gap+15
- Incentives32
- Confidence52
Docker shipped the fix in Sandboxes 0.42.0 on September 7 and published the advisory on September 15. Every macOS build from 0.28.0 through 0.41.x is affected, and the default mount is the current directory, read-write.
Reality
- Evidence70
- Adoption
- Insufficient
- Hype gap0
- Incentives65
- Confidence63
OX Security found that DeepSeek's coding harness would let an agent unlock itself simply by asking, because the gate trusted a name the caller supplies. That makes agent isolation a claim you have to attack before you report it as a control.
Reality
- Evidence70
- Adoption55
- Hype gap+18
- Incentives65
- Confidence60
A reproducible strace benchmark on a twenty-line validation task logged 752 attempted /proc/*/environ opens from Claude Code and none from Codex, which makes where an agent looks at startup a procurement question.
Reality
- Evidence60
- Adoption20
- Hype gap+26
- Incentives55
- Confidence52
A developer's build notes on giving a coding agent its own machine: scoped credentials, a snapshot before every session, and a hypervisor that boots exactly two things.
Reality
- Evidence46
- Adoption
- Insufficient
- Hype gap−6
- Incentives30
- Confidence52